src/input_agent.py

import whisper
import os
import torch
import warnings
import gc  # Garbage Collector for memory management
import tempfile
from pydub import AudioSegment # 🟢 NEW: Add pydub

# Suppress FP16 warnings on CPU
warnings.filterwarnings("ignore", message="FP16 is not supported on CPU; using FP32 instead")

class AgentInput:
    def __init__(self, device="cpu"):
        print(f"👂 Agent 1 (Input) Online: Preparing Whisper on {device}...")
        
        # USE 'tiny' FOR CLOUD TO PREVENT EXIT CODE 137 (OOM)
        # Use 'base' only if you are running on a machine with 16GB+ RAM
        self.model_name = "tiny" 
        
        try:
            # Load the model and immediately collect garbage to free RAM
            self.model = whisper.load_model(self.model_name, device=device)
            gc.collect() 
            print(f"✅ Whisper '{self.model_name}' model loaded. RAM optimized.")
        except Exception as e:
            print(f"⚠️ Load failed: {e}. Attempting emergency load...")
            # Emergency fallback to tiny if not already tried
            self.model = whisper.load_model("tiny", device=device)

    # 🟢 NEW HELPER: Sanitizes corrupted browser audio
    def _sanitize_audio(self, audio_path):
        try:
            # Try to load it regardless of format
            audio = AudioSegment.from_file(audio_path)
            # Export it to a clean, standard WAV in a temp file
            temp_path = os.path.join(tempfile.gettempdir(), f"clean_audio_{os.path.basename(audio_path)}.wav")
            audio.export(temp_path, format="wav")
            return temp_path
        except Exception as e:
            print(f"⚠️ Audio Sanitization Warning: {e}")
            return audio_path # Fallback to original if pydub fails
    
    def transcribe(self, audio_path, language=None):
        if not self.model or not audio_path: 
            return [{"text": "", "speaker": "SYSTEM"}]
        
        try:
            # 🟢 Clean the audio first!
            clean_path = self._sanitize_audio(audio_path)
            
            # Ensure fp16=False for CPU to save on conversion overhead
            result = self.model.transcribe(audio_path, language=language, fp16=False)
            
            # Clean up after transcription to keep memory low
            transcription_text = result["text"].strip()
            del result # Delete the raw result object
            gc.collect() # Force memory release

            # Clean up temp file
            if clean_path != audio_path and os.path.exists(clean_path):
                os.remove(clean_path)
                
            return [{"text": transcription_text, "speaker": "Speaker 1"}]
        except Exception as e:
            print(f"❌ Transcription Error: {e}")
            return [{"text": "", "speaker": "ERROR"}]